Command note: A review with no reply is a conversation you walked away from in public. Most accounts ask for reviews on autopilot and then never look at what came back. The request is automated. The answer usually is not, so it waits on whoever remembers. Here is the prompt we run to see the gap: "List every review this account received in the last 90 days. Group them by star rating. For each one show the source, the date, the first line of the review and whether it has a reply. Put unanswered reviews under 4 stars at the top, oldest first. Read only. Do not reply to anything." What it surfaces: • Low-star reviews sitting unanswered for weeks • Five-star reviews nobody thanked • One platform getting replies while another is ignored • The same complaint showing up more than once Run it today: 1) Pick one account and run the prompt as written 2) Reply to the oldest unanswered low-star review first, yourself 3) Thank the five-star reviews in one sitting 4) Write down any complaint that appears twice and fix the cause 5) Decide who owns review replies and how fast they answer 6) Put the prompt on the calendar for the same day every week The replies stay human. The prompt only makes sure nothing sits unseen. People read the reply before they read the rating.

Command note: A link that works for you is not a link that works for the lead. The classic miss: the message reads well, the workflow fires on time, and the button goes to a preview URL, an old page, or a login screen. You were signed in when you tested it, so it opened fine. Nobody else was. Where links quietly break: • Preview or editor URLs pasted in place of the published one • Pages renamed, moved, or unpublished after the message was written • Booking links for a calendar that has since been turned off • Links copied from one client's template into another's • Trigger links still pointed at last quarter's offer Do this before any message with a link goes live: 1) Send it to yourself. The real message, not the preview pane 2) Open it on a phone, in a private window, signed out of everything 3) Click every link, including the footer and the unsubscribe 4) Finish what the link asks for. Book the slot, submit the form 5) Confirm the record lands in the account where you expect it 6) Repeat whenever a page, calendar, or domain changes The message is not the deliverable. The click is.

Command note: Nobody reads your automation in the builder. They read it on a lock screen. Every message looks fine inside GHL. Wide editor, merge fields labeled, the whole context in your head. Your lead sees two lines of preview on a small screen, stacked between other notifications. What breaks on a phone that looked fine in the builder: • An empty merge field, so the text opens with "Hi ," • The real ask buried below the preview cut-off • A long link that eats half the screen • An email subject that truncates mid-word • A sender name the lead doesn't recognize Do this before any message goes live: 1) Send it to your own phone from the real workflow, not the preview pane 2) Read only the lock-screen preview. Is the point clear from that alone? 3) Test once with a contact missing a first name and see what renders 4) Open the email in your phone's mail app, in light and dark mode 5) Check the sender name reads like a person or business they know 6) Fix it, resend, and read it again cold When Claude drafts the copy, add one more pass: ask it to rewrite the first line so the message makes sense from the preview alone. That single edit fixes most of what the phone exposes. The builder shows what you wrote. The phone shows what they got.

Command note: An automation with no sending hours will text a new lead at 2 a.m. the moment one arrives. Most builds never set the hours. Until you turn on a Time window on each automation that sends, texts and emails can go out at any hour, including the middle of the night. What we shipped: the current GHL Command release adds a job to step 12 of Build a client account: Set the hours texts may go out. The build itself sets no sending hours, so the job tells you where to set them, and that Specific time starts at Monday to Friday, 8 to 5. Where it bites: • Form leads that land overnight • Quote follow-ups that fire on a weekend • Re-engagement sends kicked off from a late-night import • Any workflow cloned from a template that never had hours set Do it now: 1) List every workflow in the account that sends a text or email 2) Open each one and turn on its Time window 3) Pick the hours on purpose. Which hours apply to your clients is a question for your own lawyer, not a default 4) Re-check after every new build or re-plan, since a new sending workflow starts with no window 5) Write the hours on the client's handoff so they know when messages go out Speed to lead wins deals. Speed at 2 a.m. wins complaints.

I am very curious what you all are building brains for and how you currently use them, or intend to use them? One of the largest projects I've been working on for a while now is (obviously) a brain project. With Command OS soon to be released as a significant upgrade for GHL Command, one of our users, Dane, told me a couple of months ago that he wanted a brain that could oversee his GHL accounts and intervene to avoid things like an employee texting or emailing a lead something like a calendar invite, or a follow-up sales pitch, not realizing that the lead had actually taken action on that item recently. It makes us look stupid when we invite a "lead" to a sales meeting or resend them a sale nurture message, only to have them respond that they signed up 48 hours earlier. We've all been there. We've all felt stupid in that moment and thought of a witty response to cover our perceived ineptitude. That conversation gave birth to the concept of an overseeing MASTER brain that controls and interacts with smaller individual brains, each of which is assigned to a unique client's GHL sub-account, with the ability to learn from our communication styles, to read conversations, understand the business model, sales materials, products and prices, interpret customer intentions and ultimately act on our behalf. If you are an agency running client accounts, that also means learning from and acting on behalf of your clients' accounts, should you so choose. It does involve a Supabase account and lots of security and guidelines. It has taken weeks of testing, scrutiny and revisions to make sure it worked the way I wanted it to and it is ALMOST ready for deployment. In fact, it's really the only thing holding back Command OS from public release today. Today, Claude suggested we release Command OS without it and add it in the coming weeks. I refused and said it was non-negotiable. It has to be included with the initial release. We will always upgrade and improve it, but it will be included, even if that means we have to delay the initial release.